-
Jérome Perrin authored
testing with to list tests cd software/erp5/test env PYTHONNOUSERSITE=1 \ MAKEFLAGS=-j128 \ SLAPOS_SR_TEST_NAME=erp5-py3 \ SLAPOS_TEST_SKIP_SOFTWARE_REBUILD=1 \ SLAPOS_TEST_VERBOSE=0 \ SLAPOS_TEST_SKIP_SOFTWARE_CHECK=1 \ SLAPOS_TEST_DEBUG=0 \ PYTHONWARNINGS=ignore \ zope-testrunner --color -vD --path . --tests-pattern test --list-tests and to run : env PYTHONNOUSERSITE=1 \ MAKEFLAGS=-j128 \ SLAPOS_SR_TEST_NAME=erp5-py3 \ SLAPOS_TEST_SKIP_SOFTWARE_REBUILD=1 \ SLAPOS_TEST_VERBOSE=0 \ SLAPOS_TEST_SKIP_SOFTWARE_CHECK=1 \ SLAPOS_TEST_DEBUG=0 \ PYTHONWARNINGS=ignore \ zope-testrunner --color -vD --path . --tests-pattern test problems: - zope.testrunner could support setupModule / setupClass directly (or at least the workaround could be put directly in slapos.core testing on the default SlapOSInstanceTestCase) - zope.testrunner -D trigger debuggers on expected failures (probably similar to https://github.com/zopefoundation/zope.testrunner/issues/141 ) - hitting Ctrl+C seems "even worse" than default unit test -> we can probably register a signal handler in setupClass to stop slapos standalone subsystem ourselves. - nxdtest probably needs support for good reporting on test_result_module zope.testrunner
d7590fc8